1. Field of the Invention
Exemplary aspects of the present invention generally relate to an image processing device, an image forming apparatus including the image processing device, an image processing method, and an image processing program.
2. Description of the Background
Related-art image forming apparatuses, such as a copier, a facsimile machine, a printer, or a multifunction printer having two or more of copying, printing, scanning, and facsimile functions, form a toner image on a recording medium (e.g., a sheet) according to image data using an electrophotographic method. In such a method, for example, a charger charges a surface of an image bearing member (e.g., a photoconductor); an optical scanning device emits a light beam onto the charged surface of the photoconductor to form an electrostatic latent image on the photoconductor according to the image data; the electrostatic latent image is developed with a developer (e.g., a toner) to form a toner image on the photoconductor; a transfer device transfers the toner image formed on the photoconductor onto a sheet; and a fixing device applies heat and pressure to the sheet bearing the toner image to fix the toner image onto the sheet. The sheet bearing the fixed toner image is then discharged from the image forming apparatus.
In addition to a normal output mode for generating and outputting a high-quality image, some of the image forming apparatuses have a toner save mode for outputting an image with a smaller amount of an image forming material such as toner and ink to reduce consumption of the image forming material compared to that in the normal output mode. Examples of methods for reducing the consumption of the image forming material include a method for forming an image after reducing the density of an input image or obtaining a logical product of a dithered image and a decimation pattern of an input image, a method for modulating a pulse width and controlling an amount of writing light when forming an image, and so forth. However, because the image is output after reducing density in the above-described methods, the quality of the output image is degraded compared to that in the normal output mode. In general, the toner save mode is used when a higher quality image is not required, for example, when checking the layout of an output image. Nevertheless, it is still desirable to improve the quality of imaging even when the toner save mode is used.
Published unexamined Japanese Patent application No. (hereinafter referred to as JP-A-) 2001-219626 discloses a print system configured to reduce toner consumption by referring to object data, such as text, graphics, and photographs, associated with an image signal of an input image. In the print system, an image part including the text, the graphics, and the photographs in the input image is specified based on the object data of the input image to reduce an amount of toner used for a part other than the image part while not reducing an amount of toner used for the image part. As a result, a certain level of legibility of the text can be maintained even though the total amount of toner used for outputting the image is reduced. However, in the print system, the above-described processing is not performed on an input image with no object data, such as a scanner image read by an image reading device. Therefore, the print system is not applicable to an image forming apparatus such as a copier, in which the image reading device is included.
In another approach, JP-A-2005-059444 discloses an image forming apparatus configured to detect an edge part in an input image such that an amount of toner used for a non-edge part in the input image is reduced while not reducing an amount of toner used for the edge part. More specifically, an edge pixel in a scanner image read by an image reading device is detected by an edge detection unit such as a Laplacian filter so that the density of the detected edge pixel is maintained and the density of a non-edge pixel is reduced to output an image. Accordingly, in the image forming apparatus, the density of the edge part is kept relatively high even though the total amount of toner used for outputting the image is reduced, so that a certain level of legibility of text is maintained by emphasizing the edge part of the text. Moreover, the problem of the print system disclosed in JP-A-2001-219626 described above can be prevented.
However, in the image forming apparatus, the above-described processing is also performed on an edge part included in an image without text, such as a photographic image. Consequently, the density of the edge part becomes relatively high while the density of a non-edge part becomes relatively low even in the image with no text. After a study performed by the inventors of the present invention, it has been found out that unnatural or odd parts may be included in the image with no text output from the image forming apparatus depending on contents of the output image.
The inventors of the present invention have studied types of images which tend to include unnatural or odd parts when consumption of an image forming material such as toner is reduced. More specifically, consumption of the image forming material is not reduced for the edge part while reducing that for the non-edge part when outputting various types of images in a similar way as the processing performed in the image forming apparatus disclosed in JP-A-2005-059444. The output images are evaluated by multiple evaluators to determine whether or not natural or odd parts are included therein.
Results of the evaluation are illustrated in FIG. 1.
After the evaluation, the inventors of the present invention have determined that a degree of unnaturalness in output images found by the multiple evaluators differs depending on the contents of the output images. For example, all of the multiple evaluators found odd parts in an image including a human face. Specifically, a higher degree of unnaturalness is found in an output image of a face image area including a face image and a surrounding image when the density of an edge part is high and that of a non-edge part is low. A degree of unnaturalness is prominent particularly in an image including a face of an acquaintance. The reason is that outlines of hair and pupils are detected as the edge part due to having higher contrast with a background image so that density thereof becomes relatively high. By contrast, a skin portion is not detected as the edge part due to having lower contrast with the background image so that density thereof becomes relatively low. Consequently, only hair and pupils are emphasized, resulting in a higher degree of unnaturalness.
On the other hand, an output image including man-made objects such as buildings and machinery has a lower degree of unnaturalness even when the density of the edge part is relatively high and that of the non-edge part is relatively low. An output image including nature such as landscape and plants also has a higher degree of unnaturalness as compared to the output image including the man-made object, although the degree of unnaturalness is lower than that in the output image of a human face.
A specific image area including a face image area and an image area having images of nature such as landscape and plants has unnatural or odd parts when the density thereof is partially reduced as well as when the density of the non-edge part is reduced while that of the edge part is not reduced.
When the edge part is detected by an edge detection filter such as a Laplacian filter as in the case of the image forming apparatus disclosed in JP-A-2005-059444, an outline of a character image having a complicated figure and a small character image may not be detected as an edge part. In addition, when contrast between a character image and an adjacent image is low, the outline of the character image may not be detected as an edge part. Particularly, for example, in a map image in which a very small character image indicating an address and a name is included and the contrast between the character image and an adjacent image is low, it is very difficult to detect the outline of the character image as an edge part. Consequently, density of the outline of a character image including a complicated figure, a small character image, and a character image having a lower contrast with the adjacent image is reduced as well as that of parts adjacent to the above-described character images even when the processing disclosed in JP-A-2005-059444 is performed, thereby degrading legibility of the character images. Particularly when the text includes a numeric image, which is unlike a character image which indicates a word of a text, it is very difficult to identify a character indicated by the numeric image by referring to the surrounding text before and after the numeric image. Moreover, in general, a numeric character in an image often includes important data. Therefore, improvement in legibility of the numeric image of which outline may not be detected as an edge part is highly desired.